Upload job.sh with huggingface_hub

#2
Files changed (1) hide show
  1. job.sh +4 -4
job.sh CHANGED
@@ -36,9 +36,11 @@ echo "PYTORCH_CUDA_ALLOC_CONF=${PYTORCH_CUDA_ALLOC_CONF}"
36
  echo "CKPT_LOCAL_DIR=${CKPT_LOCAL_DIR} TMPDIR=${TMPDIR}"
37
 
38
  ARGS=(
 
39
  --model-size "${MODEL_SIZE:-1.7b}"
40
- --dataset "${DATASET:-HuggingFaceFW/fineweb}"
41
- --dataset-config "${DATASET_CONFIG:-sample-10BT}"
 
42
  --split "${SPLIT:-train}"
43
  --text-field "${TEXT_FIELD:-text}"
44
  --block-size "${BLOCK_SIZE:-4096}"
@@ -53,7 +55,6 @@ ARGS=(
53
  --save-interval "${SAVE_INTERVAL:-100}"
54
  --weights-save-interval "${WEIGHTS_SAVE_INTERVAL:-0}"
55
  --ckpt-local-dir "${CKPT_LOCAL_DIR}"
56
- --bucket-sync-interval "${BUCKET_SYNC_INTERVAL:-1}"
57
  --log-interval "${LOG_INTERVAL:-10}"
58
  --dtype auto
59
  --tokenizer "${TOKENIZER:-meta-llama/Llama-3.2-1B}"
@@ -66,7 +67,6 @@ ARGS=(
66
  )
67
 
68
  if [ -n "${ASYNC_BUCKET_URI:-${BUCKET_URI:-}}" ] && [ "${ASYNC_BUCKET_UPLOAD:-1}" = "1" ]; then
69
- ARGS+=(--async-bucket-uri "${ASYNC_BUCKET_URI:-${BUCKET_URI}}")
70
  fi
71
 
72
  if [ -n "${INIT_FROM:-}" ]; then
 
36
  echo "CKPT_LOCAL_DIR=${CKPT_LOCAL_DIR} TMPDIR=${TMPDIR}"
37
 
38
  ARGS=(
39
+ --sft
40
  --model-size "${MODEL_SIZE:-1.7b}"
41
+ --datasets "${DATASETS:-}"
42
+ --dataset-configs "${DATASET_CONFIGS:-}"
43
+ --dataset-weights "${DATASET_WEIGHTS:-}"
44
  --split "${SPLIT:-train}"
45
  --text-field "${TEXT_FIELD:-text}"
46
  --block-size "${BLOCK_SIZE:-4096}"
 
55
  --save-interval "${SAVE_INTERVAL:-100}"
56
  --weights-save-interval "${WEIGHTS_SAVE_INTERVAL:-0}"
57
  --ckpt-local-dir "${CKPT_LOCAL_DIR}"
 
58
  --log-interval "${LOG_INTERVAL:-10}"
59
  --dtype auto
60
  --tokenizer "${TOKENIZER:-meta-llama/Llama-3.2-1B}"
 
67
  )
68
 
69
  if [ -n "${ASYNC_BUCKET_URI:-${BUCKET_URI:-}}" ] && [ "${ASYNC_BUCKET_UPLOAD:-1}" = "1" ]; then
 
70
  fi
71
 
72
  if [ -n "${INIT_FROM:-}" ]; then